Short answer: a product page is ready for AI shopping when it has server-rendered Product and Offer schema (price, currency, availability), a product identifier (GTIN, MPN, or SKU), AggregateRating for reviews, clear returns and shipping, a real description and specs, and a robots.txt that lets AI crawlers read the page. Fix the schema and crawler items first; they decide whether an agent can see you at all.

Think of this in three layers. First, can an AI crawler read the page at all? Second, can it understand and price the product? Third, does it trust the listing enough to recommend it? A gap in the first layer makes everything below it moot, which is why the order here matters. Work top to bottom.

Layer 1 · Can AI read it

Crawler access and rendering

  • AI crawlers allowed in robots.txt. GPTBot, OAI-SearchBot, ClaudeBot, and PerplexityBot are not disallowed. Block one and that engine can never recommend you.
  • Schema is server-rendered. Your Product JSON-LD is in the raw HTML, not injected by JavaScript after load. Many AI crawlers do not run JavaScript, so client-rendered schema is invisible to them even though it looks fine to you.
  • Page returns real content to a plain request. No bot-challenge or empty shell when fetched without running scripts.

This layer is where most well-built stores fail without knowing it. The page looks perfect to a human, so nobody checks whether a non-JavaScript crawler sees the same thing.

Layer 2 · Can AI understand and price it

Product identity and data

  • Product schema present with name, image, description, and brand.
  • Offer with price and currency. An agent will not recommend a product it cannot price.
  • Availability set and accurate. Stale or missing availability gets you dropped, because an agent will not risk sending a buyer to an unclear listing.
  • Product identifier: a GTIN, MPN, or SKU so AI can match your item across the web and merge it into its product graph.
  • Real description and specs. Size, material, and compatibility answer the buyer questions that trigger a recommendation.
Layer 3 · Does AI trust it

Reviews and merchant trust

  • AggregateRating schema with a rating value and review count, the structured social proof agents lean on.
  • Visible, authentic reviews that mention real use cases, so an agent can judge whether the product delivers.
  • Organization schema so AI recognizes you as a real brand entity.
  • Clear return and shipping terms, ideally machine-readable in the Offer, so the agent trusts the merchant.
Priority order: if you fix nothing else, fix Layer 1. Server-rendered schema and open crawler access are what get you into the running at all. Layers 2 and 3 decide whether you win the shortlist once you are in it, but they cannot help a page an AI cannot read.

Run this whole checklist free, in seconds

Our free AI Shopping Readiness Checker reads a product URL and grades every item above: schema and rendering, price and availability, a GTIN, reviews, trust signals, and crawler access. It flags client-rendered schema, gives you paste-ready JSON-LD for the gaps, and even lets you compare against a competitor. No signup, no AI tokens, nothing stored.

Check my product free →

Most of these gaps map to a fix you already have a tool for. Missing reviews and AggregateRating? PopBoost renders visible, schema-backed rating widgets. Stale availability is really an inventory problem, and EZStock keeps stock and availability accurate so your in-stock signal stays true. No offers or bundles with clean pricing? EZBundle.

For the how-to behind each layer, see the guide on getting recommended in ChatGPT Shopping and the deep dive on product schema for AI shopping, which includes the exact JSON-LD to paste. If you are new to the whole shift, what agentic commerce means for a Shopify store explains why this checklist matters now.

Frequently asked questions

What is the fastest way to check AI shopping readiness? Run a free checker on one product URL. It reports schema, price, availability, identifiers, reviews, trust, and crawler access in seconds, which is far faster than inspecting each by hand.

My page looks great but scores low. Why? Almost always because the schema is client-rendered. Your browser runs the JavaScript that builds it, so you see a full page, but a non-JavaScript AI crawler sees an empty one. Move the schema into the server-rendered HTML.

Does this replace normal SEO? No. Keep your standard SEO for the classic results. AI shopping readiness is a parallel layer for the agents, and the two share a lot of the same structured-data groundwork.